This Bugzilla instance is a read-only archive of historic NetBeans bug reports. To report a bug in NetBeans please follow the project's instructions for reporting issues.
build 20080330 - create a sequence diagram - create a couple of lifelines - create a synch message from 1st lifeline to the 2nd on the right - create an asynch message from 1st lifeline to the 2nd on the right - select the 1st lifeline, now position the mouse pointer to the bottom of the 1st lifeline and select a message, eg, sych or asynch, to create, and move the mouse pointer to the bottom of the 2nd lifeline on the right and release the mouse button. Before releasing the mouse pointer, the message arrow is snapping on the previous message (asynch message) to make a nested message to the asynch message to line up horizonally - if release mouse button like this, the new message is created as nested in the asynch messsage. See the 1st attached screenshot. Result: a new lifeline is created for the new message Another scenario. - create a sequence diagram - create a couple of lifelines - create a synch message from 1st lifeline to the 2nd on the right - create a new lifeline to the left of the 1st 2 lifelines (see the 2nd attached screenshot) - from the left most (new) lifeline, create a message from there to the lifeline in the middle by positioning the mouse pointer to the top of the timeline from this new lifeline to the bottom of the middle lifeline as seen from the 2nd attached screenshot. Result: a new lifeline is created to the right of the middle lifeline to connect the new message to.
Created attachment 59444 [details] incorrect message created
Created attachment 59445 [details] incorrect message created 2
both are not a bugs, first place is not valid target so it just the same as empty space and in this case new element is created, second is real empty space of cause it can be enhanced, for first: may be temporary link can be red in case it over lifeline but place isn't correct, or some cross on target side can be drawn for second it's possible to use align like functionality and snap to lifeline even if cursor is below actual lifeline in my opinion issues are different, so need to have separate id, and may be not issues but enhancements
will try to implement some solution in nearest stabilization
suggested: same lifeline lengths + decoration to show not valid target/source + corresponding handling(if target exist but is not valid do not create new element and new message).
same lenth is implemented, it's time to report issues if it fails in some cases
http://hg.netbeans.org/uml-meteora/rev/081e0dc1b686 now source anchor in message creation isn't fixed and followw mouse, if source point isn't valid source stay in last valid position (before was suggested to move but show some sign for invalid source point).
both fixes do not cover two initial issue but was discussed as features which make such cases less likely to appear. it still possible to move target pointer below "target" lifeline and got new lifeline created and still possible to point to not valid lifeline point and got new lifeline created.
http://hg.netbeans.org/uml-meteora/rev/9d8ba86467da now if target isn't valid, for example second create message to the same lifeline, crossed red circle will be shown and no lifeline will be created on mouse release, it should solve first case, but can't verify because can't build now
because there are a lot of cases discussed in this issue and initially descibed at least two issues not sure if I can mark it as fixed, but second case which may require ability to draw new message behing existend bounds of lifeline will not be implemented in this milestone, and may be in this release(at least I prefer not to spend time now), may be second one can be separated and this one can be marked as fixed.
changed my mind, let it be fixed, please file an enhancement request for ability to draw messages below actual lifeline size, and separate new issues if some lifelines will have different lengths. opinions welcomed, if any ideas to reopen this one issue, please specify one problem per issue.
verified in build 20080512. The new implementation is very smart that it automatically moves the source anchor to line up with the target and I think that's what it should be. I see no enhancement request for now.